Get started

Model requirements

Choose a Koeva transcription model and review the practical Windows, memory, storage, CPU, and GPU guidance for that model.

A stylized graphics card and processor surrounded by an audio waveform ribbon

Baseline requirements

Koeva performs speech recognition locally, so transcription speed depends on your computer and on the model selected during installation or downloaded later in Settings. The table below separates fixed requirements from model-specific recommendations.

Component	Baseline
Operating system	Windows 10 version 1809 or later, or Windows 11, 64-bit
Storage	SSD, with free space based on the selected model
Processor	Recent 4-core CPU minimum; larger models benefit from more cores
Graphics	Optional; Vulkan acceleration is used when available and CPU fallback is automatic
Graphics driver	A current manufacturer driver is recommended when using GPU acceleration

Choose by model

Use the selector to compare the exact model files available in Koeva. These recommendations combine official Whisper model-size guidance, the GGML file sizes used by whisper.cpp, and Koeva's practical headroom for Windows, shared components, and long recordings.

They are not a benchmark guarantee. Laptop power limits, drivers, CPU fallback, audio duration, and other applications can change real processing speed.

Model selector

Choose a model, then check the hardware.

Recommended default for strong consumer PCs.

Accuracy: Very good
Speed: Fast
Model file: 834 MB

Memory: 16 GB RAM
Processor: Recent 4-core / 8-thread CPU or better
Graphics: Optional; dedicated NVIDIA, AMD, or Intel Vulkan GPU with 6 GB+ VRAM recommended. CPU fallback is slower.
Storage: Keep 5 GB free for Koeva, shared components, and working space.

This is Koeva's default balance of speed and quality.
Choose it first when the computer has a current dedicated GPU and enough graphics memory.

Reference sources

OpenAI Whisper model sizes and reference VRAM table whisper.cpp quantization guidance OpenAI Whisper large-v3-turbo model card ggerganov/whisper.cpp GGML model repository

Model comparison

The compact table below is useful when comparing several computers or deciding which model to install first.

Model	File	Memory	Graphics recommendation	Best fit
Base q8	78 MB	8 GB RAM	Optional. Vulkan acceleration is used when available; otherwise Koeva runs automatically on CPU.	Older or modest PCs, quick drafts, short dictations.
Small q8	252 MB	8 GB RAM	Optional. A Vulkan GPU improves speed; CPU fallback is automatic.	Everyday dictation on light laptops and entry desktop PCs.
Large v3 Turbo q5	547 MB	16 GB RAM	Optional; dedicated Vulkan GPU with 4 GB+ VRAM recommended. CPU fallback is slower.	Balanced quality on mid-range machines.
Large v3 Turbo q8	834 MB	16 GB RAM	Optional; dedicated NVIDIA, AMD, or Intel Vulkan GPU with 6 GB+ VRAM recommended. CPU fallback is slower.	Recommended default for strong consumer PCs.
Large v3 q5	1.01 GB	16 GB RAM minimum; 32 GB preferred for long files	Optional; dedicated Vulkan GPU with 8 GB+ VRAM recommended. CPU fallback can be substantially slower.	Higher accuracy on powerful desktops and recent gaming laptops.
Large v3	2.88 GB	32 GB RAM recommended	Optional; dedicated Vulkan GPU with 10 GB+ VRAM recommended. CPU fallback can be substantially slower.	Best quality on high-end desktops or workstation-class laptops.

Performance expectations

The exact speed varies with recording length, selected model, graphics driver, CPU/GPU backend, and laptop power settings. Review the model and GPU guidance page before choosing hardware for Koeva.